Muntin Research · Field report

How should an independent operator price a menu when every input keeps moving?

An independent restaurant prices its menu against inputs that never sit still. This practitioner field report documents Muntin's per-ingredient pricing instrument — 100 ingredients read into four postures — and grounds each analytical layer in published research: menu-cost and price-stickiness theory, farm-to-retail price transmission, edible-yield and meat-science literature, seasonal price-variation research, cross-price and commodity co-movement economics, and food-waste return studies. The document has paper form (abstract, cited sections, methods, limitations). It is not peer-reviewed, it is not a controlled experiment, and it makes no claim of statistical significance.

The headline reads are descriptive and drawn from public wholesale references, not any operator's delivered price. Of 100 ingredients, 37 lock, 19 cushion, 7 float, and 37 withhold. The worst trim tax is citrus at 2.16x and the gentlest is mushroom at 1.14x. Fifty-four ingredients carry a seasonal window that cleared a noise gate; the rest show only scatter. Across 432 large-move episodes, 94% traveled with at least one other ingredient — so a backup must be checked pair by pair, since most named substitutes still hedge and only a minority mirror; and the median episode cleared in 77 days, which is the case against reprinting a menu on a spike.

Keywords: menu pricing · edible yield · price stickiness · seasonal price gate · commodity co-movement · cross-price elasticity · food-waste return · restaurant cost control

Open the interactive playbook

Ask this paper

Seven questions this field report answers — each answer carries its own caveat so it stays true quoted short.

How should a restaurant price a menu against inflation?

Price ingredient by ingredient against each one's own baseline, not the menu as a single block. In this field report's 100 ingredients the split is 37 lock, 19 cushion, 7 float, and 37 withhold — and the instrument declines to publish a read for the 37 it can't stand behind rather than guess. This is a descriptive read of a tracked wholesale reference against each ingredient's own history, never a forecast of where a price goes next.

Grounded in: [1]

Should I reprint the menu when a price spikes?

Usually not on the spike itself. In the tracked record these large moves cleared in a median of about 77 days, with the middle half between 50 and 105 days — long enough to unsettle a menu but short enough that a reflexive reprint often reverses. That is a description of what happened in the record, not a prediction of the next spike.

Grounded in: [1]

How do I price on yield instead of the invoice?

Divide the purchase price by the edible yield: the trim tax is 1 ÷ yield, so a 45%-yield lemon costs 2.22× its purchase price per usable pound. On a category basis citrus is the steepest at 2.16× and mushroom the gentlest at 1.14×, but that is a reference multiplier, not your measured yield and not a delivered cost — individual ingredients vary around their category figure (lime is 2.86×, lemon 2.22×).

Grounded in: [14]

Does swapping to a cheaper ingredient actually save money?

Check it pair by pair — a cheaper stand-in only protects cost if it does not move with the thing it replaces. Across 432 large-move episodes in the tracked record, 94% traveled with at least one other ingredient, so most “backups” rise at the same time. Co-movement here is association across the record, not a measured cause, and it is why a swap has to be verified rather than assumed.

From the tracked record

Is it worth buying an ingredient in its cheaper month?

Sometimes — 54 of the 100 ingredients carry a seasonal window that cleared a dispersion noise gate, and the other 46 show only scatter, where a “cheapest month” would be noise dressed as a signal. Where a window cleared, buying in it is a real saving against that ingredient's own high; where it didn't, the calendar won't help you. These windows describe the tracked record, not a forecast, and “in season” is not the same as “cheapest.”

From the tracked record

What share of restaurants actually fail in year one?

About 26% in the first year — not the 90% that circulates as folklore. The widely repeated 90% figure was traced to a misquote with no data behind it; the measured first-year failure rate for independent restaurants is roughly a quarter. This is Muntin's compiled reading of source [6], a descriptive statistic, not a prediction about any one restaurant.

Grounded in: [6]

Does ingredient cost even matter to the plate?

Yes — a small farm share is not a small stake. The farm's value share of the U.S. food dollar is about 15.9¢ (a value proxy, not a volume share), and reducing kitchen waste returns roughly 7:1, not the mis-applied 14:1. A thin-margin plate can't absorb a costing mistake, which is exactly why yield and posture discipline — not the raw commodity price — move the margin. Muntin's compiled reading of sources [2] and [34].

Grounded in: [2] [34]

Our contribution

We join public wholesale-price history with open yield, seasonality, and co-movement data into a per-ingredient operator instrument — 100 ingredients read against a 134-ingredient yield reference — and add two honest gates: a seasonal noise gate that admits only windows surviving a dispersion noise gate, and a co-movement test that checks whether a culinary substitute would actually protect cost rather than mirror the problem. We then translate the food science for the line. We run no new experiments; the instrument is descriptive and reference-based. Plainly and honestly: this is not peer-reviewed, it is not a controlled study, and no result here is "significant" in the statistical sense. Every number is a co-occurrence read against a public reference, never a forecast and never a measured cause.

The problem: pricing against inputs that never sit still

Changing a posted price is not free. Mankiw showed that even a small fixed cost of repricing makes it privately rational for a firm to leave its price unchanged after a shock — and his point is the opposite of the popular reading. The cost is not the printing bill; it is that a small repricing friction has large consequences. On the line, every time you reconsider a price you spend real, non-trivial attention: guest expectation, staff retraining, the coordination of a change. Change prices deliberately, not reflexively.

The input you watch is a minority slice, and it does not pass through cleanly. USDA ERS food-dollar accounting puts the farm share at about 15.9 cents of each dollar spent on domestically produced food — correcting the folk belief that the commodity price is most of what a plate costs. Peltzman, across 242 markets, found output prices rise faster than they fall: when your input eases, retail historically does not fall as fast as it climbed. Meyer and von Cramon-Taubadel's survey shows this farm-wholesale-retail asymmetry is common and does not by itself prove collusion. The lesson for the operator: a wholesale index is a reference, not your delivered cost, and it reaches the plate slowly and unevenly.

Margins are thin, and the failure story is overstated. The National Restaurant Association's 2025 operator benchmark shows full-service median pre-tax margins are low, against the lay picture of a high-margin cash business. Parsa's Dun & Bradstreet study found first-year independent-restaurant failure near 26% — not the widely repeated 90%, which he traced to a misquote with no data behind it. The follow-on survival analysis shows failure is structured by affiliation, location, and size, not random luck. So you carry little cushion for a pricing mistake, but the odds are not the doom figure, and pricing discipline is a lever you actually hold. This instrument reads 100 ingredients through exactly that lens.

Grounded in: [1] [2] [3] [4] [5] [6] [7]

Print or float: the posture decision

Holding a price for months is normal, not lazy. Bils and Klenow measured a median price spell of about 4.3 months including temporary sales. Nakamura and Steinsson re-examined the same class of data and showed that, once sales are stripped out, regular prices change only about half as often — correcting the reading that prices are basically flexible. So a menu price you have left alone for a season is behaving the way disciplined firms behave. You are not behind for leaving it be.

Menu engineering tells you which prices are worth defending. Kasavana and Smith plot every item on two axes — menu mix (popularity) and contribution-margin dollars — not food-cost percentage, correcting the older rule that the lowest food-cost-percentage item is the most profitable. On the line that means: know which items carry your gross-profit dollars, because that is where a posture decision actually earns or costs you.

Our four postures across the 100 ingredients: 37 lock, 19 cushion, 7 float, 37 withhold. Lock means the read is clear enough to hold a price with confidence. Float — only 7 — means the input moves enough, and cleanly enough against its own baseline, to move with it. Withhold — 37 — means we decline to recommend because the public reference strays too far from delivered cost to say anything honest. Cushion — 19 — is the tighter band that has not proven out: the signal is not strong enough to lock, so we add a buffer rather than dress imprecision up as certainty. Cushion is the honest middle, not a hedge in disguise.

Proteins skew hard toward withhold. Of the 27 proteins, 16 withhold, 9 lock, 2 cushion, and none float. Proteins are where a wholesale reference wanders furthest from your invoice, so the instrument deliberately says less about them, and says it plainly.

Grounded in: [8] [9] [10]

Price on what reaches the plate: the trim tax

The unit that matters is the edible portion, not the as-purchased pound. Dopson and Hayes codify the AP/EP framework and the yield test: the test returns a yield percentage, and edible-portion cost equals as-purchased cost divided by that yield. Stated as a multiplier, the trim tax is 1 divided by edible-yield. The standardized yield tables underneath are the USDA Food Buying Guide and Agriculture Handbook No. 102 — and AH-102 is the actual primary source that most uncredited "edible-yield %" charts trace back to. Buy by the pound; cost by the plate.

Our two ends of the range: the worst-trim category is citrus at 2.16x — you buy about 2.16 units to plate one, though for citrus that surviving unit is juice, not trimmed flesh, so the figure is a juice-extraction cost rather than knife trim — and the gentlest is mushroom at 1.14x. Twenty-four of the 100 ingredients carry all four joined layers at once — a printable posture, an edible-yield cost, a cheapest month, and a checked swap. A deeper served-pound layer, where cooking loss and purge stack on raw knife trim, is measured for the 21 with a cooked yield; the yield reference underneath spans 134 ingredients.

Cooking removes weight past the knife. The USDA cooking-yield tables (Roseland and colleagues) show measured loss is dominated by moisture and fat, correcting the belief that AP-to-EP loss is mostly trimming. Tornberg's meat-protein work explains the mechanism: the hotter you take a cut, the more water its proteins wring out, so cooked yield falls with doneness. Purslow shows the shrink is not one flat percentage but climbs in steps as the center moves from medium-rare toward well-done. Weatherly and colleagues show that cutting your own steaks from a subprimal trades one cost for another. Okomoda's fish work shows that with whole fish the species, not the sticker price, decides how much you plate. The same invoice price hides very different plate costs.

For dried staples the tax runs backward. Turhan and colleagues found dried chickpeas roughly double in weight when soaked; Yadav and Jindal found milled rice comes back roughly two-and-a-half to three times. So for dried legumes and grains, yield is a gain rather than a loss — cost them hydrated, not dry.

Grounded in: [11] [12] [13] [14] [15] [16] [17] [18] [19] [20]

The hidden season and the noise gate

A window has to be separated from scatter before it counts. Cleveland and colleagues' STL procedure splits a price series into trend, a seasonal component, and a remainder, using outlier-resistant loess so a few odd weeks do not dominate. Findley and colleagues, documenting X-12-ARIMA, add identifiable-seasonality tests that correct the assumption that any recurring dip or peak is truly seasonal. On the line: a dip you saw once is not a window; a window has to survive a test.

Seasonality is item-specific, not a blanket. USDA ERS, using 2016-18 monthly prices, found fresh-vegetable seasonality is category- and item-specific — correcting the tidy story that all produce is cheapest in summer. So each ingredient earns its own window, or it earns none.

Our gate admits 54 of the 100 ingredients: they carry a seasonal timing window that cleared the noise test, and the other 46 show only scatter. Some proteins trough after their supply peak, but that window is a descriptive read of past price, not a forecast of this year. Two honesty checks come from the flavor literature. Schwieterman and colleagues note that "in season" and "cheapest" are not the same weeks. Kader shows the best-eating window is set by variety and ripeness at picking, not by price. Our window points at a historical price trough; it does not promise the trough repeats, and it does not promise flavor.

Grounded in: [21] [22] [23] [24] [25]

The swap that helps versus the swap that mirrors

Substitution is real, but it is not one number. Deaton and Muellbauer's Almost Ideal Demand System models how budget shares respond to relative prices; Okrent and Alston estimated those cross-price responses across 43 disaggregated food products. Andreyeva and colleagues, reviewing 160 studies, found no single food elasticity — correcting the shorthand that "food demand is inelastic" as one figure. So a swap can genuinely relieve a cost, but the size of the relief depends on the specific pair.

A swap that moves in step with your problem does not hedge it. Pindyck and Rotemberg documented that largely unrelated commodities show persistent co-movement well beyond anything shared fundamentals explain — correcting the myth that co-movement reveals a real linkage, or that swapping into a co-moving good protects you. Our co-movement read makes the same point on the shelf: across 432 large-move episodes, 94% traveled with at least one other ingredient. Because co-movement is that pervasive, a backup cannot be assumed independent, so it pays to check: most named substitutes still move on their own clock and do hedge, and only a minority mirror. Co-occurrence is not cause, and it is not a hedge.

Flavor is not fungible, which caps every swap. Grosch's omission work shows a dish's signature smell rests on a few specific molecules — remove them and it is gone. Potter and Fagerson show cilantro's aldehydes are not carried by any other common herb, so that swap is never 1:1. On the line the order is: the co-movement test tells you whether a swap even protects cost; the flavor chemistry tells you what it costs the plate.

Grounded in: [26] [27] [28] [29] [30] [31]

Don't reprint on a spike: the shock-duration read

Because repricing carries a cost, a transitory shock should not move the menu. Mankiw's model makes leaving the price unchanged after a shock the rational choice when changing it is costly. Blinder's interviews with about 200 firms sharpen the picture: most reported repricing no more than once or twice a year, and they ranked the physical cost of changing prices low — the real friction is customer relationships and coordination, not printing. A spike, on its own, is not a reason to reprint.

Most spikes clear on their own. Our duration read across 432 large-move episodes puts the median clearing at 77 days, with the middle half between 50 and 105 days. In operator terms, the typical spike is gone in roughly a quarter, and a menu change made in week one is often stale by the time the ink dries.

And a spike is not a forecast. Working showed that an intertemporal price spread reflects the cost of carrying stock, not the market predicting future scarcity — correcting the instinct to read a gap as a warning about where prices are headed. So read a spike as a level against the ingredient's own baseline, descriptively, and let the duration read decide whether to wait it out.

Grounded in: [1] [32] [33]

Waste is margin: trim-to-value

Cutting waste pays, and the payback is not capital-heavy. Champions 12.3's restaurant study, across 114 sites in 12 countries, found nearly every site earned a positive return, with the average restaurant saving about 7 dollars for every 1 dollar invested — correcting the belief that kitchen-waste programs demand heavy capital and long paybacks. The trim you costed as the trim tax is also the waste you can recover.

Quote the right number. The widely repeated 14:1 return comes from Champions 12.3's earlier multi-sector work, spanning manufacturing, retail, hospitality, and foodservice, and it is mis-applied when passed off as the restaurant figure. The restaurant-specific number is closer to 7:1, and that is the one we use.

Prevention beats disposal. ReFED's national analysis identified 27 cost-effective measures and found prevention — waste tracking chief among them — outranks composting and recycling. On the line the order follows: measure first, because the cheapest pound of food is the one you never trimmed into the bin.

Grounded in: [34] [35] [36]

The evidence, on the page

The finding / how-it-grounds / myth cells are Muntin's own compiled analysis (CC BY 4.0); the title, authors, and DOI are public fact. Each summary is bounded to what the source shows.

Of 100 ingredients: 37 lock, 19 cushion, 7 float, 37 withhold. The 37 withholds split 27 with no public series (band zero) and 10 too volatile to anchor (bands 30.1%–62.4%).

Category trim multipliers (134-ingredient yield reference)

Category averages — individual ingredients vary around their category figure (e.g. lime 2.86×, lemon 2.22×).
SourceSource: the 134-ingredient edible-yield reference. Trim tax = 1 ÷ mean category yield; a reference multiplier, not a delivered cost.
The evidence, on the page
#SourceLayerFindingHow it groundsMyth it correctsConf.Cited in
1N. Gregory Mankiw (1985). Small Menu Costs and Large Business Cycles: A Macroeconomic Model of Monopoly. DOImenu-costsIn a monopoly model, a small fixed cost of changing the posted price makes it privately rational for a firm to leave its price unchanged after a nominal demand shock, yet the resulting price rigidity produces welfare losses far larger than the menu cost itself.Our print-vs-float posture rests directly on Mankiw's result that a small fixed repricing cost rationally justifies holding a listed price through a shock; we carry it from macro monopoly theory into the concrete menu-reprint decision.The popular reading that a 'menu cost' is merely the trivial expense of reprinting menus — Mankiw's point is the opposite: even small repricing frictions (a metaphor for all adjustment costs) generate disproportionately large effects, so the cost is neither trivial nor merely physical.high2
2U.S. Department of Agriculture, Economic Research Service (ERS) (2023). Food Dollar Series (farm share of the U.S. food dollar; companion: Price Spreads from Farm to Consumer). DOIwhy-it-mattersIn the U.S. food-dollar accounting, farm establishments received about 15.9 cents of each dollar U.S. consumers spent on domestically produced food in 2023 (the 'farm share'), with the large majority going to post-farm marketing costs — processing, packaging, transportation, wholesaling, and retailing; the companion Price Spreads from Farm to Consumer series (a distinct ERS data product) reports farm value, wholesale, and retail values whose spreads move partly independently of farm-gate prices.Our 'wholesale is not what you pay' honesty rests directly on this official accounting: because the farm/commodity value is a minority (~16 cents) of the retail food dollar, a commodity or wholesale move cannot be read one-to-one into an operator's menu cost.Corrects the folk belief that the farm/commodity (or wholesale) price is most of what you pay and that farmers capture most of the food dollar — the 2023 farm share is about 15.9 cents; the majority is post-farm marketing cost.high1
3Sam Peltzman (2000). Prices Rise Faster than They Fall. DOIbandsAcross 242 markets (77 consumer and 165 producer goods), output prices respond faster to input-cost increases than to decreases in more than two of every three markets, with the immediate response to a positive cost shock at least twice the response to an equal negative shock and the gap persisting five to eight months — the foundational cross-market demonstration of asymmetric 'rockets and feathers' pass-through.Our wholesale-reference bands are framed as market context rather than a live retail price precisely because Peltzman shows output prices track input costs up fast but down slowly, so a wholesale dip has not necessarily flowed through to what an operator actually pays.Corrects the assumption that when wholesale or input costs fall, retail prices fall just as fast as they rose — Peltzman documents the opposite (rise-faster-than-fall) in roughly two-thirds of markets.high1
4Jochen Meyer and Stephan von Cramon-Taubadel (2004). Asymmetric Price Transmission: A Survey. DOIbandsThis agricultural-economics survey classifies the types, causes, and econometric measurement of asymmetric price transmission along farm-wholesale-retail chains, documenting that asymmetry is widespread empirically but arises from multiple mechanisms (adjustment/menu costs, inventory holding, search costs, and market power) rather than any single cause, and that measured asymmetry is sensitive to model specification and data frequency.We treat the farm-to-wholesale-to-retail chain as loosely and asymmetrically coupled per this survey, which is why our bands report a wholesale reference range rather than imputing a retail price from a wholesale move.Corrects the myth that asymmetric pass-through is proof of seller collusion or retail market power — the survey shows asymmetry also arises from non-collusive mechanisms (menu/adjustment costs, inventory, search) and appears even in competitive settings.high1
5National Restaurant Association (2025). 2025 Restaurant Operations Report (Restaurant Operations Data Abstract). DOImenu-costsDrawing on financial and operational data from more than 900 U.S. restaurants, the report shows fullservice restaurants had a median income before taxes of 2.8% of sales and limited-service restaurants 4.0%, documenting the industry's thin, low-single-digit net-margin structure.Our 'why cost discipline matters' layer rests on these low-single-digit median pre-tax margins: because the cushion is only a few points of sales, a small drift in food cost can erase an independent's entire profit, which is exactly why per-ingredient tracking matters.The lay perception that restaurants are high-margin cash businesses; the authoritative operator benchmark shows median pre-tax margins in the low single digits (2.8% fullservice / 4.0% limited-service).med1
6H. G. Parsa, John T. Self, David Njite, Tiffany King (2005). Why Restaurants Fail. DOIwhy-it-mattersUsing Dun & Bradstreet longitudinal records for independent restaurants (Columbus, Ohio, 1996-1999), first-year failure was 26.16%, and cumulative three-year failure was roughly 57-61% (franchise chains 57.2%, independents 61.4%; the three-year cumulative rate did not exceed ~60%). A literature review found no empirical basis for the widely repeated 90%-fail-in-year-one claim, and failure was also traced to non-financial drivers, notably family life-cycle / quality-of-life management and, qualitatively, location.Our analysis stands on the corrected ~26% first-year figure (not 90%) to argue honestly that restaurant failure is real but survivable, so margin and per-ingredient cost control is a lever operators genuinely hold rather than a lost cause.The 'about 90% of restaurants fail in their first year' claim: Parsa found it widely accepted with no data behind it and traced it to an American Express-linked source (the AmEx-sponsored reality show 'The Restaurant') that, when asked, stated in writing it could not provide supporting data. The peer-reviewed first-year rate for independents was ~26% (26.16%).high1
7H. G. Parsa, John Self, Sandra Sydnor-Busso, Hae Jin Yoon (2011). Why Restaurants Fail? Part II - The Impact of Affiliation, Location, and Size on Restaurant Failures: Results from a Survival Analysis. DOIwhy-it-mattersA survival (hazard) analysis of restaurant failures showing that mortality is systematically structured by affiliation (chain/multi-unit vs. independent), location (zip code), and size rather than random: independents faced substantially elevated closure risk versus chain-affiliated restaurants, and smaller restaurants failed at higher rates than larger ones.We extend this by treating the finding that independents carry structurally higher failure hazard as the reason their thinner cash buffers make disciplined per-ingredient costing a survival tool, not a nicety.—high1
8Mark Bils and Peter J. Klenow (2004). Some Evidence on the Importance of Sticky Prices. DOIpostureUsing BLS micro-data on ~350 categories (about 70% of consumer spending), the median price spell lasts about 4.3 months including temporary sales and about 5.5 months once sales are excluded, with the frequency of price change differing dramatically across goods.Gives our 'don't reprint on day one' posture an empirical cadence: even in a broad, relatively flexible sample, half of prices persist 4-5+ months — long enough that a short cost shock usually resolves before a disciplined operator would otherwise reprice.—high1
9Emi Nakamura and Jón Steinsson (2008). Five Facts about Prices: A Reevaluation of Menu Cost Models. DOIpostureRe-examining the same class of BLS micro-data, once temporary sales are excluded the median frequency of regular (nonsale) price change is roughly half the with-sales frequency, implying regular prices persist on the order of 8-11 months, and the price-change hazard does not rise with time since the last change.Our strongest anchor for holding through a shock: stripping out promo sales shows regular list prices persist most of a year, so a 77-day median shock will typically pass before a reprint is warranted — the empirical spine of 'don't reprint on day one.'The widely repeated claim that Bils-Klenow prove prices 'change every ~4 months, so prices are basically flexible' — Nakamura-Steinsson show most of that churn is temporary sales snapping back to a sticky regular price, so regular prices are roughly twice as sticky (median ~8-11 months).high1
10Michael L. Kasavana; Donald I. Smith (1982). Menu Engineering: A Practical Guide to Menu Analysis. OCLC 9550099 — https://search.worldcat.org/title/9550099 (ISBN 0-932235-01-5 / 9780932235015; also Open Library OL21382210M)postureIntroduced menu engineering: every item is plotted on two axes — menu mix (popularity) and contribution margin (gross profit dollars per item) — into Stars (popular, high-margin), Plowhorses (popular, low-margin), Puzzles (unpopular, high-margin), and Dogs (unpopular, low-margin), each carrying a distinct reprice/reposition/keep/cut action.Our per-item posture verdict (hold / reprice / reposition / cut) is the direct descendant of their popularity x contribution-margin quadrant; we extend it by driving the margin axis with live cost-index movement instead of a static food cost.Corrects the older 'lowest food-cost-percentage item is the most profitable' rule: menu engineering ranks on contribution-margin dollars, not food-cost %, so a low-cost-% item can still be a weak contributor and a high-cost-% item a strong one.high1
11Dopson, Lea R.; Hayes, David K. (2019). Food and Beverage Cost Control, 7th Edition (Ch. 5, Monitoring Food and Beverage Product Costs). DOItrim-taxCodifies the AP/EP framework and the yield test (butcher's/cooking test): the test produces a yield percentage, and edible-portion (servable) cost equals the as-purchased cost divided by that yield percentage — i.e., a cost factor of 1 divided by yield applied to the AP price.This is the canonical textbook statement of our trim-tax formula; our layer applies its cost factor (1 / yield %) as the multiplier on ingredient cost when costing and pricing a menu item.—high1
12U.S. Department of Agriculture, Food and Nutrition Service (2024). Food Buying Guide for Child Nutrition Programs (interactive web-based edition). DOItrim-taxPublishes standardized as-purchased (AP) to served/edible-portion yield data in a six-column table (Column 1 Food As Purchased; Column 3 Servings per Purchase Unit, EP; Column 5 Purchase Units for 100 Servings; Column 6 Additional Yield Information) so an operator can calculate the AP quantity to buy to net a required amount of served food. It is the USDA's standardized, continuously-updated purchase-yield reference for Child Nutrition Programs, and is widely used beyond schools as a government foodservice purchase-yield reference.Our trim-tax multiplier (1 / edible yield) uses FBG-style AP-to-served yield factors as its standardized, citable basis; we extend the FBG's purchase-quantity math into a per-dollar cost penalty applied when pricing a menu item.Corrects the assumption that the FBG is only for school/CACFP kitchens: its title scopes it to Child Nutrition Programs, but the underlying AP-to-served yield tables are a general USDA purchase-yield reference usable by any foodservice operator.high1
13Pecot, Rebecca K.; Watt, Bernice K. (U.S. Department of Agriculture, Agricultural Research Service) (1956). Food Yields Summarized by Different Stages of Preparation (Agriculture Handbook No. 102). DOItrim-taxCompiles empirical percentage yields for a wide range of foods across successive preparation stages (as-purchased to trimmed/edible to cooked), quantifying refuse/trim losses and cooking losses relative to as-purchased weight.It is the primary empirical foundation the trim-tax rests on — documenting that edible yield after trim and cooking is materially below purchased weight; we convert those measured loss fractions into the 1/yield cost markup.Many circulating 'edible-yield %' charts have no cited provenance; AH-102 is the actual primary USDA source those yield figures trace back to, not vendor lore.high1
14Roseland, Janet M.; Nguyen, Quynh Anh; Williams, Juhi R.; Patterson, Kristine Y.; Showell, Bethany A.; Pehrsson, Pamela R. (USDA Agricultural Research Service) (2017). USDA Table of Cooking Yields for Meat and Poultry, Release 2. DOItrim-taxReports measured cooking yields (weight change from moisture and fat loss) for specific meat and poultry cuts under standardized cooking protocols, isolating cooking loss as a distinct component of edible yield separate from knife trim.Supplies the cooking-loss component of edible yield for proteins specifically; we use it to justify splitting the trim-tax into a trim component and a cook component rather than a single blended factor.Corrects the belief that AP-to-EP yield loss is mostly knife trimming: for cooked proteins a large share of weight loss is moisture and fat driven off during cooking, not trim.high1
15Tornberg, E. (2005). Effects of heat on meat proteins – Implications on structure and quality of meat products. DOIcooked-yieldA review of how heat restructures meat proteins. The globular sarcoplasmic and myofibrillar proteins begin to denature and aggregate around 40°C, with most set by about 60°C and coagulation continuing up to ~90°C; as this network denatures it shrinks and squeezes out held water, so cooking (water) loss rises with temperature. Collagen (connective tissue) denatures at roughly 53-63°C and its fibers shrink, and on continued moist heating the shrunken collagen dissolves into gelatin unless it is locked up by heat-stable crosslinks.——high1
16Purslow, P.P.; Oiseth, S.; Hughes, J.; Warner, R.D. (2016). The structural basis of cooking loss in beef: Variations with temperature and ageing. DOIcooked-yieldIn aged beef eye-of-round (semitendinosus), the water squeezed out during cooking comes mainly from the muscle fibers themselves shrinking, not from connective tissue. Fibers shrink sideways (cross-sectional area down about 20-24%) as myosin denatures around 50-65 C, then shrink lengthwise as actin denatures around 70-75 C — so cooking loss climbs in two temperature-driven steps as the center gets hotter. Meat aged longer (14 days vs 1 day) gave up more fluid when cooked.——high1
17Weatherly, B.H.; Griffin, D.B.; Johnson, H.K.; Walter, J.P.; De La Zerda, M.J.; Tipton, N.C.; Savell, J.W. (2001). Foodservice yield and fabrication times for beef as influenced by purchasing options and merchandising styles. DOItrim-taxProfessional cutters at three foodservice purveyors fabricated beef subprimals into ready-to-cook portion cuts while researchers weighed every cut and timed every job. Both the total usable yield and the labor time depended on the cutting spec, and the two moved against each other: as the target steak size got smaller, more steaks came off a given subprimal, but total foodservice yield fell and total cutting time rose — an inverse relationship between how much usable product you kept and how long the cutting took.——med1
18Okomoda, V.T.; Solomon, S.G.; Wukatda, S.S.; Ikape, S.I.; Ikhwanuddin, M.; Abol-Munafi, A.B. (2021). Fillet Yield and Length-Weight Relationship of Five Fish Species From Lower River Benue, Makurdi, Nigeria. DOItrim-taxFilleting 60 fish each of five species, researchers found edible/fillet yield swung widely by species: three species (Mormyrus rume, Labeo senegalensis, and the catfish Clarias gariepinus) yielded at least 55% edible parts, while the other two yielded 39% or less. For most species the yield percentage stayed roughly constant regardless of fish size; only the catfish yield rose with larger fish.——high1
19Turhan, M.; Sayar, S.; Gunasekaran, S. (2002). Application of Peleg model to study water absorption in chickpea during soaking. DOIcooked-yieldSoaking dried chickpeas follows a predictable curve: water rushes in fast at first, then the uptake rate tapers as the seed nears saturation. The authors showed the two-constant Peleg equation reproduces this entire soaking curve — for five winter- and five spring-planted chickpea genotypes across water temperatures from 20 to 100 C — using only early-timepoint data, and that warmer soak water hydrates faster (the model's equilibrium form estimates the fully-hydrated endpoint at temperatures at or above 40 C). A soaked-out chickpea has taken on roughly its own dry weight in water.——high1
20Yadav, B. K.; Jindal, V. K. (2007). Water uptake and solid loss during cooking of milled rice (Oryza sativa L.) in relation to its physicochemical properties. DOIcooked-yieldCooking 10 Thai milled-rice varieties (16-29% amylose) in excess water, the authors found water uptake climbs quickly at first then flattens as grains approach saturation (uptake curves fit a modified exponential, R2 0.995-0.999), while dissolved solids leach into the cooking water as a power function of cooking time — steadily more the longer it boils. How much water a rice takes up was governed largely by its amylose content and amylose solubility (alkali-spreading score); higher-amylose rices take up more water and need longer cooking.——high1
21Robert B. Cleveland, William S. Cleveland, Jean E. McRae, and Irma Terpenning (1990). STL: A Seasonal-Trend Decomposition Procedure Based on Loess (with Discussion). DOIseasonal-windowCleveland, Cleveland, McRae and Terpenning introduced STL, a robust loess-based procedure that splits a time series into trend, a seasonal component allowed to evolve gradually over time, and a remainder, giving a resistant way to estimate the recurring seasonal shape without a single anomalous observation distorting it.Our seasonal-window layer estimates each ingredient's recurring shape in the STL spirit—decomposing a price history into trend + seasonal + remainder with a robust, time-varying seasonal component—so an outlier week does not redraw the window and a slowly-shifting season is still captured.—high1
22David F. Findley, Brian C. Monsell, William R. Bell, Mark C. Otto, and Bor-Chung Chen (1998). New Capabilities and Methods of the X-12-ARIMA Seasonal-Adjustment Program. DOIseasonal-windowDocumenting the U.S. Census Bureau's X-12-ARIMA program, the authors describe its seasonal-adjustment method and diagnostic suite, including the tests used to decide whether a series' seasonal movement is statistically identifiable (the combined stable-plus-moving F-tests and Kruskal-Wallis test, after Lothian and Morry) rather than a series that merely looks seasonal.Our noise gate operationalizes the same principle these diagnostics formalize: a seasonal window is drawn only when the recurring component is large and stable relative to the remainder, so scatter is never promoted to a 'season'—we require identifiable seasonality before publishing a window.Corrects the myth that any recurring dip or peak is 'seasonal'; the identifiable-seasonality framework this program reports shows apparent patterns can fail the stable/moving/nonparametric tests and must be treated as non-seasonal noise.high1
23U.S. Department of Agriculture, Economic Research Service (2024). Starchy fresh vegetables (excluding potatoes) had the most seasonal price variation from 2016-18 (Charts of Note, from Food-at-Home Monthly Area Prices data). DOIseasonal-windowUsing Food-at-Home Monthly Area Prices for 2016-18, USDA ERS found that seasonal price variation among fresh vegetables is category- and item-specific: it is largest for the 'other starchy vegetables' group (dominated by sweet corn), which is cheapest in summer when domestic supply peaks, and smallest for dark green vegetables, which show little seasonal price movement.This grounds our per-ingredient (not blanket) seasonal windows: because seasonality varies by category and item, we estimate a window for each ingredient rather than assuming a uniform 'summer is cheap' rule, and we suppress windows for low-variation items like dark greens.Corrects the 'all produce is cheapest in summer / produce is uniformly seasonal' oversimplification; ERS shows some categories have pronounced summer troughs while others (dark greens) show minimal seasonal price variation.high1
24Schwieterman, Michael L.; Colquhoun, Thomas A.; Jaworski, Elizabeth A.; Bartoshuk, Linda M.; Gilbert, Jessica L.; Tieman, Denise M.; Clark, David G. (2014). Strawberry flavor: diverse chemical compositions, a seasonal influence, and effects on sensory perception. DOIpeak-seasonAcross many strawberry varieties and harvest dates, sweetness and overall consumer liking tracked sugar levels and a specific set of aroma volatiles — and both varied substantially with growing season and conditions, with fruit grown under warmer-season pressures generally lower in sucrose and total volatiles. Notably, a subset of volatile compounds made berries taste sweeter than their measured sugar alone would predict. So the very same cultivar can eat noticeably better in a favorable stretch of the season than in a harsh one, independent of price.——high1
25Kader, Adel A. (2008). Flavor quality of fruits and vegetables. DOIpeak-seasonA review of what actually makes fruits and vegetables taste good. Eating quality — the balance of sugars, acids, and aroma volatiles — is largely fixed before harvest by the cultivar and by how ripe the item was when it was picked. For most crops, the longer the gap between harvest and eating, the more of that characteristic flavor is lost and the more off-flavors develop. Many commercial crops are harvested less-than-ripe so they survive shipping, which caps how good they can ever taste on the plate.——high1
26Angus Deaton; John Muellbauer (1980). An Almost Ideal Demand System. DOIsubstitutionIntroduced the Almost Ideal Demand System (AIDS), a theoretically consistent demand model in which each good's budget share is linear in log prices and log real expenditure, yielding estimable own-price, cross-price, and expenditure elasticities and permitting direct tests of homogeneity and symmetry.It is the demand-system backbone our substitution finder rests on: substitutability is a property of estimated compensated demand response, so we treat 'substitute' as a demand-model claim, never a raw price-correlation claim.—high1
27Abigail M. Okrent; Julian M. Alston (2012). The Demand for Disaggregated Food-Away-from-Home and Food-at-Home Products in the United States. DOIsubstitutionEstimating a demand system for 43 disaggregated food-away-from-home and food-at-home products in the United States via two-stage budgeting, they recover own-price, expenditure, and cross-price elasticities and find many statistically significant substitution and complementary relationships within and across food categories, with food-away-from-home items being expenditure-elastic 'luxuries.'It supplies real U.S. food cross-price elasticities showing substitution is pair- and category-specific and frequently small, so our finder derives a candidate swap's hedging value from measured demand response rather than assuming any two ingredients substitute.—high1
28Tatiana Andreyeva; Michael W. Long; Kelly D. Brownell (2010). The Impact of Food Prices on Consumption: A Systematic Review of Research on the Price Elasticity of Demand for Food. DOIsubstitutionA systematic review of 160 studies found mean own-price elasticities of demand for major U.S. food and beverage categories clustering in the inelastic range (absolute values roughly 0.27 to 0.81), with food away from home, soft drinks, juice, and meats the most price-responsive (about 0.7-0.8) and eggs the least (about 0.27).It grounds the finder's category-specific humility: measured food price-responsiveness varies widely by category and robust cross-price/substitution estimates are comparatively scarce, so a proposed swap's hedging benefit must be evidenced per category rather than assumed from a single 'food is inelastic' rule of thumb.The myth of a single or uniform food price elasticity (that 'food demand is inelastic' as one number); the review documents that elasticities differ substantially by category, with food away from home among the most price-responsive.high1
29Robert S. Pindyck; Julio J. Rotemberg (1990). The Excess Co-Movement of Commodity Prices. DOIsubstitutionPrices of largely unrelated raw commodities (wheat, cotton, copper, gold, crude oil, lumber, cocoa) exhibit persistent co-movement well in excess of anything that common macroeconomic fundamentals - inflation, industrial production, interest rates, and exchange rates - can explain.It is the empirical warrant for our core honesty rule - that two ingredients whose wholesale prices move together are not thereby substitutes and swapping one for the other does not hedge - because co-movement is driven by shared macro/financial forces, not by any demand relationship between the goods (we flag that later work, e.g. Deb-Trivedi-Varangis 1996, contests the magnitude of the 'excess,' though not that unrelated commodities co-move).The myth that price co-movement reveals an economic linkage - that co-moving goods are substitutes/complements, or that a co-moving swap hedges; Pindyck-Rotemberg show commodity-price co-movement exceeds and is unexplained by common fundamentals, so co-movement is not evidence of substitutability.med1
30W. Grosch (2001). Evaluation of the Key Odorants of Foods by Dilution Experiments, Aroma Models and Omission. DOIsubstitutionAcross foods as different as coffee, boiled beef, olive oil, wine, and French fries, a food's characteristic aroma is carried by a surprisingly small set of 'key odorants.' When researchers rebuild the aroma from pure compounds (an aroma model) and then leave one compound out (an omission test), the smell noticeably shifts — direct proof that a handful of specific molecules are load-bearing for the food's identity. The review notes odorants with higher odour activity values are usually essential, with some exceptions.——high1
31T. L. Potter and I. S. Fagerson (1990). Composition of coriander leaf volatiles. DOIsubstitutionThe characteristic smell of fresh cilantro (coriander) leaf comes overwhelmingly from a family of long-chain aliphatic aldehydes — chiefly (E)-2-decenal and decanal, alongside other C10–C16 2-alkenals and n-alkanals. In this study these aldehydes were the dominant volatiles isolated from the leaf, and they are what the nose reads as 'cilantro.'——high1
32Alan S. Blinder, Elie R. D. Canetti, David E. Lebow, and Jeremy B. Rudd (1998). Asking About Prices: A New Approach to Understanding Price Stickiness. DOImenu-costsIn structured interviews with ~200 U.S. firms testing twelve theories of price stickiness, most firms reported repricing no more than once or twice a year (many only annually), and the best-supported explanations were coordination failure (waiting for rivals to move first) and implicit contracts / not antagonizing customers rather than the physical cost of changing prices.Validates our posture with operator testimony: experienced price-setters deliberately hold prices roughly annually and cite relational/strategic frictions, which is exactly why we counsel floating quietly rather than reprinting the menu at the first cost move.That 'menu costs' are dominated by the physical cost of changing price tags or menus — Blinder's survey ranked such physical costs low; the binding frictions are informational and relational (coordination failure, customer goodwill).high1
33Holbrook Working (1949). The Theory of Price of Storage. DOIwhy-it-mattersWorking demonstrated that the price relationship between a commodity for delivery at different dates (the intertemporal or 'storage' spread) is set by the quantity of supply currently on hand and the competitively-determined cost of carrying it forward, not by the market's expectation of future price change—so the recurring harvest-glut-to-lean-season price structure is the return earned for storing supply across time.This underwrites why a seasonal window exists at all—storage arbitrage smooths but cannot erase the glut-to-lean cost cycle—while cautioning that the spread is a return to storage, not a forecast, which is exactly why our seasonal-window layer prices around a recurring shape and never forecasts a level.Corrects the common belief that a forward/seasonal price gap is the market 'predicting' future scarcity; Working shows the spread reflects current supply and carrying cost, so it is not a price forecast to be traded on.high1
34Austin Clowes, Craig Hanson, and Richard Swannell (on behalf of Champions 12.3) (2019). The Business Case for Reducing Food Loss and Waste: Restaurants. DOItrim-to-valueA study of 114 restaurant sites across 12 countries found nearly every site achieved a positive return, with the average restaurant saving about $7 for every $1 invested in cutting kitchen food waste; sites reduced waste roughly 26% within the first year, over 75% recouped their investment within a year (89% within two years), and every site kept total investment under $20,000.Our trim-to-value layer uses this restaurant-specific 7:1 as its conservative, real-world ROI anchor for turning measured trim and waste reduction into recovered margin, rather than leaning on the inflated cross-sector number.The belief that kitchen food-waste programs demand heavy capital and have long paybacks - the study found every one of the 114 sites invested under $20,000 and most recouped within a single year.high1
35Craig Hanson and Peter Mitchell (on behalf of Champions 12.3) (2017). The Business Case for Reducing Food Loss and Waste. DOIwhy-it-mattersAcross 1,200 business sites in 700 companies spanning 17 countries, nearly every site achieved a positive return on food-loss-and-waste reduction, with half realizing a benefit-cost ratio of 14:1 or greater (an average of roughly $14 saved for every $1 invested).It grounds our why-it-matters case that measuring and acting on waste reliably pays back at scale, while our page deliberately cites the restaurant-specific 7:1 instead of this headline 14:1 to avoid overstating the return for a single kitchen.The widely repeated '14:1 return on food-waste reduction' is a multi-sector average (manufacturing, retail, hospitality, foodservice) that is routinely mis-applied to restaurants specifically; the restaurant-only companion study puts the figure at about 7:1.high1
36ReFED (Rethink Food Waste Through Economics and Data), a multi-stakeholder collaboration (2016). A Roadmap to Reduce U.S. Food Waste by 20 Percent. DOItrim-to-valueReFED's first national economic analysis identified 27 cost-effective, scalable solutions that could cut US food waste about 20% (roughly 13 million tons per year) and generate about $100 billion in cumulative economic value over a decade, finding prevention far more economically valuable per ton than recycling and naming Waste Tracking & Analytics the single most economically valuable business-facing solution (an estimated $1.3 billion in gross annual business cost savings), with restaurants and foodservice positioned to capture the large majority of that prevention profit potential.It grounds the trim-to-value mechanism - that the highest-ROI lever is measuring and tracking waste before cutting it, and that restaurants and foodservice capture most of the resulting profit - which our layer frames as measure-then-trim.The assumption that composting and recycling are the top food-waste levers - ReFED found prevention solutions such as waste tracking deliver far greater economic value per ton than recycling food scraps.high1

Methods

The instrument reads 100 ingredients (16 of them recently added staples) against public wholesale reference prices — never an operator's delivered price. Each ingredient is sorted into one of four postures: lock, cushion, float, or withhold (37, 19, 7, and 37 respectively). The 37 withholds are two mechanically opposite populations: 27 have no public wholesale series to read at all — most herbs, all shellfish, the whole fish, a few produce — so their band is zero, the tightest possible, because there is nothing to measure; the other 10 do carry a series, and a wide one, with bands from 30.1% to 62.4% against their own baseline, withheld for swinging too far past the 30% float line to anchor rather than for any lack of signal. The trim tax is computed as 1 divided by edible-yield, with yields taken from published tables and cost-control texts rather than our own bench measurement; 24 of the 100 carry all four joined layers at once (posture, edible cost, cheapest month, and a checked swap); a deeper served-pound layer adds cooking loss and purge on top of raw trim for the 21 with a measured cooked yield, drawing on a 134-ingredient yield reference. Category trim multipliers run from citrus at 2.16x (worst) to mushroom at 1.14x (gentlest); for citrus the edible figure is juice yield, so its multiplier is a juice-extraction cost, not knife trim. Seasonal windows are admitted only when the cheapest month clears a dispersion noise gate — its median must beat the dearest month's own 25th-percentile week, and the peak-to-trough swing must exceed the ordinary within-month spread; 54 of the 100 carry a window that clears it, and the rest read as scatter. Co-movement is read across 432 large-move episodes; 94% traveled with at least one other ingredient and only 6% moved alone, and it is this pervasiveness of co-movement that the substitute-hedge test turns on. Episode duration is the time a price stays elevated against the ingredient's own baseline window: median 77 days, 25th-to-75th percentile 50 to 105 days. Every read is descriptive and based on co-occurrence.

How sure we are

How sure we are, layer by layer. Two of these reads are close to definitions, and we lean on them hardest: the trim tax is one divided by edible yield, arithmetic, not an estimate; and the shock-duration figures are the median and quartiles of 432 measured episode lengths (77 days, half of them between 50 and 105). The seasonal windows and the substitute-mirror flags are lighter — they rest on a handful of episodes per ingredient and on the cutoffs below, so they point rather than prove — and a posture sitting near a band edge is a close call. Where the price-theory literature appears — menu costs, price stickiness, the excess co-movement of commodities — it explains why an operator behaves as they do; it is context we stand on, not a measurement of our own numbers.

Limitations

Name them plainly. The yields are reference and textbook figures, not measurements from our own kitchen, so any single lot, cut, or supplier can differ from the table. The price history is a public wholesale reference, not delivered cost — proteins diverge from it most, which is why 16 of the 27 proteins withhold rather than recommend. Co-movement is co-occurrence, never causation: the substitute test reports only whether two prices tended to move together, not why, and a driver attached to a mover is an association, not a measured cause. Seasonal windows are descriptive reads of past price, not forecasts, and "in season" is not the same as "cheapest" or "best-eating." This is a single-operator practitioner study with paper form: it is not peer-reviewed, it is not a controlled experiment, and no result is statistically significant. All figures shift as the underlying data refreshes.

Sample size per ingredient is small, and we would rather you know it. The detector keeps only each ingredient’s six largest sustained moves, so every per-ingredient read — its co-movement and its named substitute-mirror — rests on at most six of that item’s own episodes, and usually fewer. The mirror flag can fire on as little as one shared episode out of two, because it asks only that a companion move with the ingredient in at least half of those few episodes. Read a single named mirror as a flag to check on your own invoices, not a proof.

The cutoffs are chosen, not derived. A band inside 8% of its baseline earns a lock, inside 20% a cushion, inside 30% a float, and wider than 30% withholds — though width is only one of the two ways an ingredient lands in withhold; a cheapest month is named only if it beats its dearest by at least 15%; a lock must also clear 60% proven coverage. These lines are reasonable and applied consistently, but they are our lines — nudge them and a handful of ingredients cross from one call to the next. Treat any posture sitting near a threshold as a close call, not a verdict.

The reads are in-sample, and the set is self-selected. Every band, window, and duration is measured on the same multi-year record it is then scored against; there is no held-out test that a past pattern must hold again, and nothing here is offered as a guess about a future price. Only ingredients with enough committed history earn a confident band at all, and the 37 the instrument withholds are not one kind of uncertainty but two: 27 have no public wholesale series to read — herbs, shellfish, whole fish — so the instrument is blind, not undecided; the other 10 read loud and clean but swing 30.1% to 62.4% against their own baseline, too wide to anchor a printed price. A withheld ingredient is not a stable one; it is one we will not guess at.

Data availability

The full per-ingredient table behind every posture, band, trim tax, and cheapest month is a public download — 100 ingredients, released CC BY 4.0: menu-pricing.json (structured) and menu-pricing.csv (spreadsheet). Every band is a wholesale reference read against each ingredient's own baseline window — never a delivered or invoice price; the trim tax is 1 divided by edible-yield; a cheapest month is named only when the seasonal trough clears the dispersion noise gate, and a blank means the ingredient priced year-round. Each column maps to the Methods above; every read is descriptive of the tracked record and based on co-occurrence, never a forecast and never a measured cause.

Price the plate, not the invoice; hold the menu through a spike, which across our 432 episodes cleared in a median of about 77 days; and swap only when the co-movement test says the substitute actually protects cost — not when it merely moves in step with the ingredient you are trying to escape.

References

  1. 1N. Gregory Mankiw (1985). Small Menu Costs and Large Business Cycles: A Macroeconomic Model of Monopoly. The Quarterly Journal of Economics, vol. 100, no. 2, pp. 529-538 (MIT Press at time of publication; QJE now published by Oxford University Press). https://doi.org/10.2307/1885395
  2. 2U.S. Department of Agriculture, Economic Research Service (ERS) (2023). Food Dollar Series (farm share of the U.S. food dollar; companion: Price Spreads from Farm to Consumer). USDA Economic Research Service (ERS) data product. https://www.ers.usda.gov/data-products/food-dollar-series
  3. 3Sam Peltzman (2000). Prices Rise Faster than They Fall. Journal of Political Economy (University of Chicago Press), 108(3), 466-502. https://doi.org/10.1086/262126
  4. 4Jochen Meyer and Stephan von Cramon-Taubadel (2004). Asymmetric Price Transmission: A Survey. Journal of Agricultural Economics (Wiley), 55(3), 581-611. https://doi.org/10.1111/j.1477-9552.2004.tb00116.x
  5. 5National Restaurant Association (2025). 2025 Restaurant Operations Report (Restaurant Operations Data Abstract). National Restaurant Association (industry benchmark report). https://restaurant.org/
  6. 6H. G. Parsa, John T. Self, David Njite, Tiffany King (2005). Why Restaurants Fail. Cornell Hotel and Restaurant Administration Quarterly, 46(3), 304-322 (SAGE). https://doi.org/10.1177/0010880405275598
  7. 7H. G. Parsa, John Self, Sandra Sydnor-Busso, Hae Jin Yoon (2011). Why Restaurants Fail? Part II - The Impact of Affiliation, Location, and Size on Restaurant Failures: Results from a Survival Analysis. Journal of Foodservice Business Research, 14(4), 360-379 (Taylor & Francis). https://doi.org/10.1080/15378020.2011.625824
  8. 8Mark Bils and Peter J. Klenow (2004). Some Evidence on the Importance of Sticky Prices. Journal of Political Economy, vol. 112, no. 5, pp. 947-985 (University of Chicago Press). https://doi.org/10.1086/422559
  9. 9Emi Nakamura and Jón Steinsson (2008). Five Facts about Prices: A Reevaluation of Menu Cost Models. The Quarterly Journal of Economics, vol. 123, no. 4, pp. 1415-1464 (MIT Press at time of publication; QJE now published by Oxford University Press). https://doi.org/10.1162/qjec.2008.123.4.1415
  10. 10Michael L. Kasavana; Donald I. Smith (1982). Menu Engineering: A Practical Guide to Menu Analysis. Hospitality Publications, Okemos, MI (book; framework developed at the Michigan State University School of Hospitality Business). OCLC 9550099 — https://search.worldcat.org/title/9550099 (ISBN 0-932235-01-5 / 9780932235015; also Open Library OL21382210M)
  11. 11Dopson, Lea R.; Hayes, David K. (2019). Food and Beverage Cost Control, 7th Edition (Ch. 5, Monitoring Food and Beverage Product Costs). Wiley (ISBN 9781119524991). https://www.wiley.com/en-us/food-and-beverage-cost-control-7th-edition-p-9781119524991
  12. 12U.S. Department of Agriculture, Food and Nutrition Service (2024). Food Buying Guide for Child Nutrition Programs (interactive web-based edition). USDA Food and Nutrition Service — continuously updated online resource (updated for the 2024 final rule aligning Child Nutrition meal patterns with the 2020-2025 Dietary Guidelines). https://www.fns.usda.gov/tn/fbg
  13. 13Pecot, Rebecca K.; Watt, Bernice K. (U.S. Department of Agriculture, Agricultural Research Service) (1956). Food Yields Summarized by Different Stages of Preparation (Agriculture Handbook No. 102). U.S. Department of Agriculture, Agriculture Handbook No. 102 (orig. 1956; rev. 1975). https://www.ars.usda.gov/ARSUserFiles/80400530/pdf/ah102.pdf
  14. 14Roseland, Janet M.; Nguyen, Quynh Anh; Williams, Juhi R.; Patterson, Kristine Y.; Showell, Bethany A.; Pehrsson, Pamela R. (USDA Agricultural Research Service) (2017). USDA Table of Cooking Yields for Meat and Poultry, Release 2. USDA Agricultural Research Service — Ag Data Commons dataset. https://doi.org/10.15482/USDA.ADC/1409031
  15. 15Tornberg, E. (2005). Effects of heat on meat proteins – Implications on structure and quality of meat products. Meat Science. https://doi.org/10.1016/j.meatsci.2004.11.021The hotter you take a cut, the more water its proteins wring out and the less it weighs on the plate — that is your cooked yield falling as doneness climbs, so a well-done steak gives up clearly more moisture than one pulled at medium-rare. Connective-tissue-heavy cuts are the opposite bet: hold them hot and wet long enough and the collagen that made them tough melts to gelatin, which is exactly why a chuck or shank must be cooked far past the point that would ruin a loin.
  16. 16Purslow, P.P.; Oiseth, S.; Hughes, J.; Warner, R.D. (2016). The structural basis of cooking loss in beef: Variations with temperature and ageing. Food Research International, 89(Part 1), 739-748. https://doi.org/10.1016/j.foodres.2016.09.010Cooking shrink is not one flat percentage — it is driven by how hot you take the center, in steps. Pushing a steak from medium-rare up toward well-done crosses a second wave of water loss, leaving a smaller, lighter portion on the plate; well-done tickets literally shrink more product than rare ones. If you spec a cooked or served weight, the doneness you cook to changes your yield and your cost per plate. And a longer-aged cut, though more tender, can shed a bit more moisture in the pan — worth knowing when you plate to a target weight.
  17. 17Weatherly, B.H.; Griffin, D.B.; Johnson, H.K.; Walter, J.P.; De La Zerda, M.J.; Tipton, N.C.; Savell, J.W. (2001). Foodservice yield and fabrication times for beef as influenced by purchasing options and merchandising styles. Journal of Animal Science, 79(12), 3052-3061. https://doi.org/10.2527/2001.79123052xWhen you buy a subprimal (a whole strip loin, ribeye, or tenderloin) and cut your own steaks, two costs push against you at once: trimming closer and portioning smaller loses more to trim AND burns more butcher minutes. The low per-pound subprimal price is not your real cost — figure true plate cost on the yielded, cut weight plus the labor clock. A looser portion spec, or buying one step closer to portion-ready, can pencil out cheaper than full in-house breakdown once labor is counted.
  18. 18Okomoda, V.T.; Solomon, S.G.; Wukatda, S.S.; Ikape, S.I.; Ikhwanuddin, M.; Abol-Munafi, A.B. (2021). Fillet Yield and Length-Weight Relationship of Five Fish Species From Lower River Benue, Makurdi, Nigeria. Tropical Life Sciences Research, 32(1), 163-174. https://doi.org/10.21315/tlsr2021.32.1.10With whole fish, the species — not the sticker price — decides how much you actually plate. Two whole fish at the same price per pound can hand you very different edible pounds: one might fillet out at over half its weight, another under 40%, so the 'cheaper' fish can cost more per usable ounce once you account for head, frame, and skin. Always convert whole-fish AP price to a per-fillet-pound (EP) cost using that species' real dress-out — and since yield mostly does not change with fish size, one yield figure per species holds across the sizes you buy.
  19. 19Turhan, M.; Sayar, S.; Gunasekaran, S. (2002). Application of Peleg model to study water absorption in chickpea during soaking. Journal of Food Engineering, 53(2), 153-159. https://doi.org/10.1016/S0260-8774(01)00152-2Dried chickpeas roughly double in weight by the time they're soaked, so a pound of dry garbanzos becomes about two pounds hydrated before they ever hit the pot — that swell is your real yield behind the low dry price on the invoice. Most of the water goes in during the first few hours, so an overnight soak is plenty; you gain little past that. If you're behind, a hot soak gets you fully hydrated faster than a cold one.
  20. 20Yadav, B. K.; Jindal, V. K. (2007). Water uptake and solid loss during cooking of milled rice (Oryza sativa L.) in relation to its physicochemical properties. Journal of Food Engineering, 80(1), 46-54. https://doi.org/10.1016/j.jfoodeng.2006.06.021Rice swells because the grain drinks water until it's saturated — that's your cooked yield (dry rice comes back roughly two-and-a-half to three times its weight cooked). But boil it loose in lots of water for too long and starch bleeds out into the pot: you lose yield and get gluey, cloudy water. Match the water to the rice — a firm long-grain (high amylose) wants more water and time than a softer short-grain — and pull it at absorption rather than drown-and-drain if you care about finished weight.
  21. 21Robert B. Cleveland, William S. Cleveland, Jean E. McRae, and Irma Terpenning (1990). STL: A Seasonal-Trend Decomposition Procedure Based on Loess (with Discussion). Journal of Official Statistics (Statistics Sweden), vol. 6, no. 1, pp. 3-73. https://www.semanticscholar.org/paper/STL-:-A-Seasonal-Trend-Decomposition-Procedure-on-Cleveland/585bf445ec84c1d9621b2726bdcce9f544b515c8
  22. 22David F. Findley, Brian C. Monsell, William R. Bell, Mark C. Otto, and Bor-Chung Chen (1998). New Capabilities and Methods of the X-12-ARIMA Seasonal-Adjustment Program. Journal of Business & Economic Statistics (American Statistical Association / Taylor & Francis), vol. 16, no. 2, pp. 127-152. https://doi.org/10.1080/07350015.1998.10524743
  23. 23U.S. Department of Agriculture, Economic Research Service (2024). Starchy fresh vegetables (excluding potatoes) had the most seasonal price variation from 2016-18 (Charts of Note, from Food-at-Home Monthly Area Prices data). USDA Economic Research Service. https://www.ers.usda.gov/data-products/charts-of-note/chart-detail?chartId=109127
  24. 24Schwieterman, Michael L.; Colquhoun, Thomas A.; Jaworski, Elizabeth A.; Bartoshuk, Linda M.; Gilbert, Jessica L.; Tieman, Denise M.; Clark, David G. (2014). Strawberry flavor: diverse chemical compositions, a seasonal influence, and effects on sensory perception. PLOS ONE, 9(2), e88446. https://doi.org/10.1371/journal.pone.0088446'In season' and 'cheapest' are not the same weeks. Berries from the cool front and tail of the local season often eat sweeter and more aromatic than the heat-grown flood-of-supply peak, even at similar or higher cost. Taste before you commit a menu feature — a mid-heatwave flat can be watery and flat even when it's abundant and cheap. Build berry desserts and garnishes around the genuine peak-eating window, and in the off weeks lean on macerating (sugar + acid) or cooking to compensate rather than plating them raw and naked.
  25. 25Kader, Adel A. (2008). Flavor quality of fruits and vegetables. Journal of the Science of Food and Agriculture, 88(11), 1863-1868. https://doi.org/10.1002/jsfa.3293The best-eating window is decided before the crate hits your dock — by variety and by ripeness at picking, not by anything you do in the kitchen. Two levers you actually control: buy each item in its true local season (when it's picked riper and travels less), and shorten the clock from delivery to plate. A tomato or peach that spent a week in transit has already spent its flavor; hold produce cold and turn it fast instead of 'letting it ripen' warm on a shelf, which for most items just ages it rather than sweetening it.
  26. 26Angus Deaton; John Muellbauer (1980). An Almost Ideal Demand System. American Economic Review (American Economic Association), 70(3): 312-326. https://www.jstor.org/stable/1805222
  27. 27Abigail M. Okrent; Julian M. Alston (2012). The Demand for Disaggregated Food-Away-from-Home and Food-at-Home Products in the United States. USDA Economic Research Service, Economic Research Report No. 139 (ERR-139). https://www.ers.usda.gov/publications/pub-details?pubid=45006
  28. 28Tatiana Andreyeva; Michael W. Long; Kelly D. Brownell (2010). The Impact of Food Prices on Consumption: A Systematic Review of Research on the Price Elasticity of Demand for Food. American Journal of Public Health, 100(2): 216-222. https://doi.org/10.2105/AJPH.2008.151415
  29. 29Robert S. Pindyck; Julio J. Rotemberg (1990). The Excess Co-Movement of Commodity Prices. The Economic Journal (Royal Economic Society), 100(403): 1173-1189. https://academic.oup.com/ej/article-abstract/100/403/1173/5188414
  30. 30W. Grosch (2001). Evaluation of the Key Odorants of Foods by Dilution Experiments, Aroma Models and Omission. Chemical Senses, 26(5), 533–545. https://doi.org/10.1093/chemse/26.5.533A dish's signature smell hangs on a few specific molecules, not on 'roughly similar' ingredients. That's the hard, mechanistic reason a substitution is never a perfect match: if the swap doesn't carry the same key odorants, the guest's nose registers that something is off even if they can't name it. Use it as a decision rule — when you sub an ingredient, ask whether it actually delivers the one or two compounds that define the original (the sear, the citrus peel oil, the specific herb). If it doesn't, change how the dish is framed rather than pretending it's the same, and reserve the real ingredient for the plates where it's genuinely doing the work.
  31. 31T. L. Potter and I. S. Fagerson (1990). Composition of coriander leaf volatiles. Journal of Agricultural and Food Chemistry, 38(11), 2054–2056. https://doi.org/10.1021/jf00101a011Cilantro's flavor lives in a specific set of aldehyde molecules that no other common herb on your line carries in the same combination. That's why swapping in parsley or reaching for 'a little more lime' never actually reads as cilantro — you can match the green color and the fresh look, but not the aldehyde signature. If a plate is built on cilantro, treat it as non-substitutable for flavor: 86 it or re-spec the dish, don't fake it. It also explains why cilantro fades fast once chopped and heated — those volatile aldehydes cook off — so add it late and raw.
  32. 32Alan S. Blinder, Elie R. D. Canetti, David E. Lebow, and Jeremy B. Rudd (1998). Asking About Prices: A New Approach to Understanding Price Stickiness. Russell Sage Foundation, New York (ISBN 9780871541215). https://www.russellsage.org/publications/book/asking-about-prices
  33. 33Holbrook Working (1949). The Theory of Price of Storage. The American Economic Review (American Economic Association), vol. 39, no. 6, pp. 1254-1262. https://www.jstor.org/stable/1816601
  34. 34Austin Clowes, Craig Hanson, and Richard Swannell (on behalf of Champions 12.3) (2019). The Business Case for Reducing Food Loss and Waste: Restaurants. Champions 12.3, in partnership with the World Resources Institute (WRI) and WRAP. https://champions123.org/publication/business-case-reducing-food-loss-and-waste-restaurants
  35. 35Craig Hanson and Peter Mitchell (on behalf of Champions 12.3) (2017). The Business Case for Reducing Food Loss and Waste. Champions 12.3, in partnership with the World Resources Institute (WRI) and WRAP. https://champions123.org/publication/business-case-reducing-food-loss-and-waste
  36. 36ReFED (Rethink Food Waste Through Economics and Data), a multi-stakeholder collaboration (2016). A Roadmap to Reduce U.S. Food Waste by 20 Percent. ReFED. https://refed.org/downloads/Executive-Summary.pdf

Cite this & download (CC BY 4.0)

Recomputed from open data · verified 2026-07-11 · node scripts/build-study-dataset.mjs --check

This field report and its evidence table are released under CC BY 4.0 — reuse with attribution to Muntin Cost Index.

APAThe Muntin Desk. (2026). The Menu-Pricing Playbook — Muntin. Muntin Cost Index. https://muntin.digital/cost-index/menu-pricing/study/

BibTeX@techreport{muntin_menu_pricing_2026,
title = {The Menu-Pricing Playbook — Muntin},
author = {{The Muntin Desk}},
institution = {Muntin Cost Index},
type = {Field report},
year = {2026},
url = {https://muntin.digital/cost-index/menu-pricing/study/},
note = {CC BY 4.0}
}

RISTY - RPRT
TI - The Menu-Pricing Playbook — Muntin
AU - The Muntin Desk
PY - 2026
DA - 2026-07-11
PB - Muntin Cost Index
UR - https://muntin.digital/cost-index/menu-pricing/study/
LA - en
N1 - CC BY 4.0; DOI pending
ER -

CSL-JSON{"type":"report","id":"muntin-menu-pricing-2026","title":"The Menu-Pricing Playbook — Muntin","author":[{"literal":"The Muntin Desk"}],"issued":{"date-parts":[[2026,7,11]]},"publisher":"Muntin Cost Index","URL":"https://muntin.digital/cost-index/menu-pricing/study/","license":"https://creativecommons.org/licenses/by/4.0/","language":"en-US"}

study.json CC BY study.csv datapackage.json CITATION.cff datacite.json menu-pricing.json menu-pricing.csv

study.json/.csv: the evidence table — the report’s claims × the 36 sources that ground them (with DOIs). menu-pricing.json/.csv: the per-ingredient data behind the numbers. The badge means a recomputation of the reported figures from open data and a disclosed recipe — explicitly not replication of a finding.

A practitioner field report that stands on peer-reviewed work; it is not peer-reviewed and not a controlled experiment. Descriptive, never a forecast; co-occurrence, never cause; a wholesale reference, never the delivered price. Every figure is our own or attributed to its cited source.

Open the interactive playbook